The 2023 plenary lecture by Atul Butte, MD, PhD, highlighted how big data and machine learning are defining a new wave of evidence-based medicine. His view that “evidence-based medicine is data-driven medicine” and that “data is the new soil," allude to his life-long bet to leveraged data as fertile ground on which unpredictable but logical ways to improve medical practice are cultivated using artificial intelligence (AI). This is an exciting new area of health informatics and personalized medicine that might put AI on a collision path with key opinion leaders if ignored.
Trained as a computer scientist and pediatrician, Butte is a Priscilla Chan and Mark Zuckerberg distinguished professor at the University of California, San Francisco (UCSF). He is also the chief data scientist for the University of California Health Academic Centers and inaugural director of the Bakar Computational Health Sciences Institute at UCSF.
With 20 years of research funding from the NIH, 24 patents, and over 300 published papers, Butte is truly a prolific physician-scientist and inventor. He has been a member of the National Academy of Medicine since 2015 and was recognized in 2013 as a White House Champion of Change for promoting science through publicly available data. His plenary lecture this year and CV truly suggest where big data and deep learning go, big ideas are not too far behind.
His lecture illustrated how data-driven initiatives have transformed medical practice within the University of California health system. His research began with successful efforts to use machine learning along with public cancer genomics and real-life drug therapy data for drug repositioning. This approach led to a successful startup called NuMedii that focuses on finding support for new FDA-approved indications for available drugs.
Another company, Personalis, Inc., leveraged publicly available cancer genomics data and machine learning to help improve cancer diagnostics and treatment.
Butte also talked about how one of the most expansive data sets, electronic health records (EHR), could be leveraged using deep machine learning to identify safe and effective ways to cut pharmacotherapy costs or improve treatment strategies for common diseases such as diabetes.
Using a combined “modern database” composed of EHRs from nearly 9 million patients in the UC Health System, this plenary truly lived up to anticipation.
Attendees got key insights on how Internal Review Board reliance and centralized contracting can help build a single accountable care organization (ACO) and clinically integrated networks that are compatible with a deep learning healthcare system. This model of care has been adopted at the Center for Real World evidence at UCSF where AI is used to find actionable answers from available clinical data.
In the end, the lecture painted a picture of the future of medicine and how AI will shape it. With the patient always at center stage, federal legislation continues to advocate for better patient access to health records and future accountable care is likely to use AI to help health professionals and patients alike.
Large healthcare systems such as UC Health System continue to find new ways to share health records with patients, including deidentified records that can be applied to a large language model. Butte said he already can foresee a future where AI will play a role in helping patients understand their health records.
Technologies such as ChatGPT are already being tested in applications such as identification of investigational new drugs and interpretation of complex genomics. As such, one take home message was key opinion leaders might already be on a collision path with AI.